Confidence and Caveats

Client Trust Signals

Define confidence, entropy, stability, automation likelihood, lies/tamper indicators, privacy resistance, and redaction state.

Public-Safe Boundary

Local analysis works without login. Protected actions use PLATPHORM_API_KEY only.

Raw IPs, raw high-entropy component values, full visitor hashes, and unredacted JA4 digest metadata are excluded from public surfaces.

Active

Confidence combines component availability, entropy, stability, unsupported APIs, spoof indicators, and algorithm version.

Protected

Trust scoring is protected/internal-style support. Public pages explain signals without making access-control decisions.

Active

API and MCP responses emit trace headers for correlation without placing secrets or raw fingerprint payloads into traces.